Mozaic Works

Blast from the past:Microtools with LLMs

Microtools with LLMs – Mozaic Works A recent LinkedIn post by Simon Wardley raises an interesting point: what if we would build microtools fit for our environments using LLMs? Let's examin...

Blast from the past:Yes, you can test it before it's done

Yes, you can test it before it’s done – Mozaic Works A common statement we hear from testers in agile organizations is that "we can't test it until it's done". Alex breaks down in this video…

Programming language debates might soon matter as much as tabs vs. spaces.
That's the direction AI coding assistants are pushing software architecture — and it's worth taking seriously before it happens to your codebase.
As AI generates good-enough implementations, the real engineering work shifts to splitting problems into small, well-specified modules with clear contracts and tests — black boxes an agent can regenerate on demand. It's an appealing model, but it comes with real risks: duplicated bugs across modules, operational complexity we already know from microservices, and code nobody can maintain once the AI assistance is gone.

More here: https://mozaicworks.com/uncategorized/specs-over-code-how-ai-is-changing-software?fsp_sid=1959

AI assistants can prove invaluable to software architecture - if you use them to their strengths https://mozaicworks.com/blog/ai-assistants-for-software-architects-how-to-use-them-right?fsp_sid=1966